Modern server farms or datacenters typically employ a large number of servers to handle processing needs for a variety of application services. Each server handles various operations and requires a certain level of power consumption to maintain these operations. Some of these operations are “mission critical” operations, interruptions to which may lead to significant security breach or revenue losses for users associated with these operations.
One source of interruptions is failures or faults at PSUs to a server system. A failure or fault in one or more PSUs can force a sudden shutdown of a server system, possibly resulting in data losses or even damages to the server system. In order to test and improve reliability, stability and safety of PSUs, a burn-in test has been used in the production process of PSUs. However, there are many inherent problems associated with using traditional burn-in tests.